Automatic brain tumor segmentation with domain adaptation

Lutao Dai, Tengfei Li, Hai Shu, Liming Zhong, Haipeng Shen, Hongtu Zhu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Deep convolution neural networks, in particular, the encoder-decoder networks, have been extensively used in image segmentation. We develop a deep learning approach for tumor segmentation by combining a modified U-Net and its domain-adapted version (DAU-Net). We divide training samples into two domains according to preliminary segmentation results, and then equip the modified U-Net with domain adaptation structure to obtain a domain invariant feature representation. Our proposed segmentation approach is applied to the BraTS 2018 challenge for brain tumor segmentation, and achieves the mean dice score of 0.91044, 0.85057 and 0.80536 for whole tumor, tumor core and enhancing tumor, respectively, on the challenge’s validation data set.

Original languageEnglish (US)
Title of host publicationBrainlesion
Subtitle of host publicationGlioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Revised Selected Papers
EditorsMauricio Reyes, Spyridon Bakas, Theo van Walsum, Alessandro Crimi, Farahani Keyvan, Hugo Kuijf
PublisherSpringer Verlag
Pages380-392
Number of pages13
ISBN (Print)9783030117252
DOIs
StatePublished - Jan 1 2019
Event4th International MICCAI Brainlesion Workshop, BrainLes 2018 held in conjunction with the Medical Image Computing for Computer Assisted Intervention Conference, MICCAI 2018 - Granada, Spain
Duration: Sep 16 2018Sep 20 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11384 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference4th International MICCAI Brainlesion Workshop, BrainLes 2018 held in conjunction with the Medical Image Computing for Computer Assisted Intervention Conference, MICCAI 2018
CountrySpain
CityGranada
Period9/16/189/20/18

Keywords

  • Brain tumor
  • Confusion loss
  • Domain adaptation
  • Encoder-decoder network
  • Segmentation

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Automatic brain tumor segmentation with domain adaptation'. Together they form a unique fingerprint.

  • Cite this

    Dai, L., Li, T., Shu, H., Zhong, L., Shen, H., & Zhu, H. (2019). Automatic brain tumor segmentation with domain adaptation. In M. Reyes, S. Bakas, T. van Walsum, A. Crimi, F. Keyvan, & H. Kuijf (Eds.), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - 4th International Workshop, BrainLes 2018, Held in Conjunction with MICCAI 2018, Revised Selected Papers (pp. 380-392). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11384 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-11726-9_34